文件名称:GPU-speedup
介绍说明--下载内容均来自于网络,请自行研究使用
针对LA在解决大规模优化问题时需要消耗大量的时间无法达到实时性的问题,结合GPU的高速并行性,本文提出了一种基于GPU(Graphics Processing Unit,GPU)加速的细粒度并行免疫算法。本算法借助CUJDA(Compute Unifled Device Architecture,CUDA)统一架构,将实现过程转化成CUDA线程块并行计算过程,使得免疫算法在GPU中加速执行,在取得较好的优化效果的同时,解决了细粒度并行的群体规模限制问题,提高了算法的运行速度。-Solution for large-scale optimization problems in LA need to consume a lot of time when not achieve real-time issues, combined with high-speed GPU parallelism, we propose a novel GPU (Graphics Processing Unit, GPU) to accelerate the fine-grained parallel immune algorithm . The algorithm with CUJDA (Compute Unifled Device Architecture, CUDA) unified architecture, the implementation process into the process of parallel computing CUDA thread block, making the immune algorithm to accelerate the implementation of the GPU, the optimization to achieve better results in the same time, to solve the fine-grained degree of parallel group size restrictions, to improve the algorithm speed.
相关搜索: cuda
(系统自动生成,下载前可以参看下载内容)
下载文件列表
基于GPU加速的细粒度模型并行免疫算法研究.pdf